Difference between revisions of "ZIM file format"

From openZIM
Jump to navigation Jump to search
(Specified geo index)
Line 357: Line 357:


Old Zeno files used the QUnicode library instead. By switching to UTF-8 the new format is more standard-adherent and easier to understand.
Old Zeno files used the QUnicode library instead. By switching to UTF-8 the new format is more standard-adherent and easier to understand.
== Split ZIM files ==
ZIM files can be split in multiple chunks. This is necessary to be able to store big (over 4GB for example) ZIM files to limited file systems (like FAT32). That said, the chunks can be of any size, but the naming is really important. The ZIM file chunks should be named like following (the file name extensions matter): ''foobar.zimaa, foobar.zimab, foobar.zimac''...


== See also ==
== See also ==

Revision as of 12:34, 15 October 2016

Schema File Format.png

The ZIM file format is based on the old and deprecated Zeno File Format. See also a walk through example at ZIM File Example. It starts with a header, which is described here:

Header

A ZIM file starts with a header. This is offset 0.

Length in bytes, all types are little-endian.

Field Name Type Offset Length Description
magicNumber integer 0 4 Magic number to recognise the file format, must be 72173914
version integer 4 4 ZIM=5, bytes 1-2: major, bytes 3-4: minor version of the ZIM file format
uuid integer 8 16 unique id of this zim file
articleCount integer 24 4 total number of articles
clusterCount integer 28 4 total number of clusters
urlPtrPos integer 32 8 position of the directory pointerlist ordered by URL
titlePtrPos integer 40 8 position of the directory pointerlist ordered by Title
clusterPtrPos integer 48 8 position of the cluster pointer list
mimeListPos integer 56 8 position of the MIME type list (also header size)
mainPage integer 64 4 main page or 0xffffffff if no main page
layoutPage integer 68 4 layout page or 0xffffffffff if no layout page
checksumPos integer 72 8 pointer to the md5checksum of this file without the checksum itself. This points always 16 bytes before the end of the file.
geoIndexPos integer 80 8 pointer to the geo index (optional). Present if mimeListPos is at least 80.

MIME Type List (mimeListPos)

The MIME type list always follows directly after the header, so the mimeListPos also defines the end and size of the ZIM file header.

The MIME types in this list are zero terminated strings. An empty string marks the end of the MIME type list.

Field Name Type Offset Length Description
<1st MIME Type> string 0 zero terminated declaration of the <1st MIME Type>
<2nd MIME Type> string n/a zero terminated declaration of the <2nd MIME Type>
... string ... zero terminated ...
<last entry / end> string n/a zero terminated empty string - end of MIME type list

URL Pointer List (urlPtrPos)

The URL pointer list is a list of 8 byte offsets to the directory entries.

The directory entries are always ordered by URL (including the namespace prefix). Ordering is simply done by comparing the URL strings in binary.

Since directory entries have variable sizes this is needed for random access.

Field Name Type Offset Length Description
<1st URL> integer 0 8 pointer to the directory entry of <1st URL>
<2nd URL> integer 8 8 pointer to the directory entry of <2nd URL>
<nth URL> integer (n-1)*8 8 pointer to the directory entry of <nth URL>
... integer ... 8 ...

Zimlib caches directory entries and references the cached entries via the URL pointers.

Title Pointer List (titlePtrPos)

The title pointer list is a list of article indices ordered by title. The title pointer list actually points to entries in the URL pointer list. Note that the title pointers are only 4 bytes. They are not offsets in the file but article numbers. To get the offset of an article from the title pointer list, you have to look it up in the URL pointer list.

Field Name Type Offset Length Description
<1st Title> integer 0 4 pointer to the URL pointer of <1st Title>
<2nd Title> integer 4 4 pointer to the URL pointer of <2nd Title>
<nth Title> integer (n-1)*4 4 pointer to the URL pointer of <nth Title>
... integer ... 4 ...

The indirection from titles via URLs to directory entries has two reasons:

  • the pointer list is only half in size as 4 bytes are enough for each entry
  • accessing directory entries by title also makes use of cached directory entries which are referenced by the URL pointers, as implemented in zimlib.

Directory Entries

Directory entries hold the meta information about all articles, images and other objects in a ZIM file.

There are many types of directory entries:

Article Entry

Field Name Type Offset Length Description
mimetype integer 0 2 MIME type number as defined in the MIME type list
parameter len byte 2 1 (not used) length of extra paramters
namespace char 3 1 defines to which namespace this directory entry belongs
revision integer 4 4 (optional) identifies a revision of the contents of this directory entry, needed to identify updates or revisions in the original history
cluster number integer 8 4 cluster number in which the data of this directory entry is stored
blob number integer 12 4 blob number inside the compressed cluster where the contents are stored
url string 16 zero terminated string with the URL as refered in the URL pointer list
title string n/a zero terminated string with an title as refered in the Title pointer list or empty; in case it is empty, the URL is used as title
parameter data see parameter len (not used) extra parameters

Redirect Entry

Field Name Type Offset Length Description
mimetype integer 0 2 0xffff for redirect
parameter len byte 2 1 (not used) length of extra paramters
namespace char 3 1 defines to which namespace this directory entry belongs
revision integer 4 4 (optional) identifies a revision of the contents of this directory entry, needed to identify updates or revisions in the original history
redirect index integer 8 4 pointer to the directory entry of the redirect target
url string 12 zero terminated string with the URL as refered in the URL pointer list
title string n/a zero terminated string with an title as refered in the Title pointer list or empty; in case it is empty, the URL is used as title
parameter data see parameter len (not used) extra parameters

Linktarget or deleted Entry

Field Name Type Offset Length Description
mimetype integer 0 2 0xfffe for linktarget, 0xfffd for deleted entry
parameter len byte 2 1 (not used) length of extra paramters
namespace char 3 1 defines to which namespace this directory entry belongs
revision integer 4 4 (optional) identifies a revision of the contents of this directory entry, needed to identify updates or revisions in the original history
url string 16 zero terminated string with the URL as refered in the URL pointer list
title string n/a zero terminated string with an title as refered in the Title pointer list or empty; in case it is empty, the URL is used as title
parameter data see parameter len (not used) extra parameters

Cluster Pointer List (clusterPtrPos)

The cluster pointer list is a list of 8 byte offsets which point to all data clusters in a ZIM file.

Field Name Type Offset Length Description
<1st Cluster> integer 0 8 pointer to the <1st Cluster>
<1st Cluster> integer 8 8 pointer to the <2nd Cluster>
<nth Cluster> integer (n-1)*8 8 pointer to the <nth Cluster>
... integer ... 8 ...

Clusters

The clusters contain the actual data of the directory entries. Clusters can be compressed or uncompressed. The purpose of the clusters are that data of more than one directory entry can be compressed inside one cluster, making the compression much more efficient. Typically clusters have a size of about 1 MB.

The first byte of the cluster identifies if it is compressed (4) or not (0). The default is uncompressed indicated by a value of 0 or 1 (obsoleted, inherited by Zeno) while compressed clusters are indicated by a value of 4 which indicates LZMA2 compression (or more precisely XZ, since there is a XZ header). There have been other compression algorithms used before (2: zlib, 3: bzip2) which have been removed. The zimlib uses xz-utils as a C++ implementation of lzma2, for Java see XZ-Java.

To find the data of a specific directory entry within a cluster the uncompressed cluster has a list of pointers to blobs within the uncompressed cluster after the first byte.

Field Name Type Offset Length Description
compression type integer 0 1 0: default (no compression), 1: none (inherited from Zeno), 4: LZMA2 compressed
The following data bytes have to be uncompressed!
<1st Blob> integer 1 4 offset to the <1st Blob>
<2nd Blob> integer 5 4 offset to the <2nd Blob>
<nth Blob> integer (n-1)*4+1 4 offset to the <nth Blob>
... integer ... 4 ...
<last blob / end> integer n/a 4 offset to the end of the cluster
<1st Blob> data n/a n/a data of the <1st Blob>
<2nd Blob> data n/a n/a data of the <2nd Blob>
... data ... n/a ...

The offset addresses uncompressed data. The last pointer points to the end of the data area. So there is always one more offset than blobs. Since the first offset points to the start of the first data, the number of offsets can be determined by dividing this offset by 4. The size of one blob is calculated by the difference of two consecutive offsets.

Geo Index

The geo index contains a spatially indexed data structure so that it is possible to search for articles by their position on the globe. Coordinates are assumed to be WGS84 coordinates, but encoded into binary in the following way: Both latitudes and longitudes use the full span of 32 bit unsigned big endian integers, which means that the south pole has an encoded latitude of 0 and the north pole an encoded latitude of 0xffffffff (actually 0x100000000). This means that the scaling between degree and encoded integers is different for latitudes and longitudes. The coordinate of Berlin, which is approximately at 52° 31′ N, 13° 23′ E or 52.516667, 13.383333 is encoded as latitude (52.516667 + 90) / 180 * 0x100000000 = 3400580132 and longitude (13.383333 + 180) / 360 * 0x100000000 = 2307153030.

The index actually consists of a list of indices assumed to be ordered by importance. The idea is that the indices are searched in order until a certain amount of search results is found, such that at low zoom levels, only the important articles (countries, large cities) are displayed.

All offsets in the geo index are counted relative to the start of the geo index. This means that in order to get absolute file offsets, you have to add the value of the geoIndexPos field in the file header.

At geoIndexPos, the geo index starts with the following table:

Field Name Type Offset Length Description
indicesCount integer 0 4 number of geo indices (=n)
<1st offset> integer 4 4 offset to the data area of the 1st geo index
<2nd offset> integer 8 4 offset to the data area of the 2nd geo index
... integer ... 4 ...
<nth offset> integer n*4 4 offset to the data area of the nth geo index
<end offset> integer (n+1)*4 4 offset to the end of the data area of the nth geo index

Each geo index is a binary tree of coordinates alternating between latitudes and longitudes (starting with latitudes) ending in a list of coordinates and article indices as leaves. A non-leaf node is encoded as follows:

Field Name Type Offset Length Description
value integer ? 4 discriminator coordinate, cannot be zero (would indicate a leaf node)
greaterOffset integer ? + 4 4 offset to the node containing coordinates with latitude / longitude greater than the discriminator
<smaller or equal coordinates> ... ... ... node leading to coordinates with latitude / longitude smaller or equal than the discriminator
<larger coordinates> ... greaterOffset ... node leading to coordinates with latitude / longitude greater than the discriminator

Leaf nodes containing the actual coordinates and links to the articles are encoded as follows:

Field Name Type Offset Length Description
0 integer ? 4 value identifying leaf nodes
numCoordinates integer ? + 4 4 number of coordinates that follow
<1st latitude> integer ? + 8 4 1st latitude
<1st longitude> integer ? + 12 4 1st longitude
<1st title index> integer ? + 16 4 1st title index
... ... ... ... ...
<nth latitude> integer ? + n * 12 - 4 4 nth latitude
<nth longitude> integer ? + n * 12 4 nth longitude
<nth title index> integer ? + n * 12 + 4 4 nth title index

The title indices in the above table are article numbers just like in the title pointer list.


It can be assumed that the index is generated in the following way: All coordinates are sorted by latitude. The median is stored as discriminator value, the first half of the coordinates is stored in a "lower node", the second half in an upper node. For each of the two nodes, the coordinates are sorted by longitude and are split in much the same way until there are less than a specific number of articles (e.g. 10) left at which point they are stored as a list in a leaf node.

Namespaces

Namespaces seperate different types of directory entries - which might have the same title - stored in the ZIM File Format.

They can be distinguished by prepending the article namespace before the article name in the URL path, eg. http://localhost/A/Articlename.

Namespace Description
- layout, eg. the LayoutPage, CSS, favicon.png (48x48), JavaScript and images not related to the articles
A articles - see Article Format
B article meta data - see Article Format
I images, files - see Image Handling
J images, text - see Image Handling
M ZIM metadata - see Metadata
U categories, text - see Category Handling
V categories, article list - see Category Handling
W categories per article, category list - see Category Handling
X fulltext index - see ZIM Index Format

URLs

ZIM contents are addressed using URLs fitting the following pattern: <namespace>/<article_url>. The references in articles HTML code (<a href=""></a>, <img src="">, etc.) are URL-encoded following the RFC 1738 rules.

Absolute URLs, ie. with a leading slash (/) are forbidden, because this avoid including the ZIM contents in any HTTP sub-hierachy. ZIM contents URLs must consequently be relative. Be careful, <article_url> may itself contain slashes (for example "BMW_501/502").

The URLs in the UrlPointerlist are not encoded. Some readers process the requests that already do the decoding internally whereas most readers will handle the URLs directly. In this case you have to do the decoding before you pass the parameter to zimlib, but zimlib already provides a method to do so.

Local Anchors

Many articles - especially when a table of contents is used - use local anchors to jump within an article.

<a href="../A/foo#headline1">jump to article foo, headline 1</a>

The browser handles these local anchors by itself. It will determine if another article has to be loaded (local anchor inside another article than the currently shown) and will send a request only with the article URL without the local anchor - in our example "../A/foo". After the article has been loaded the browser will then search for the local anchor tag and jump to the right location.

If you use a common rendering engine or HTML widget you don't have to care for this cases, you can just use the requests as they are submitted by the engine / widget.

Should you render the article contents by yourself you have to consider this and take care of it before you hand requests to zimlib.

Encodings

Character Encoding

The standard encoding for ZIM file content is UTF-8. So both article data and URLs should be handled accordingly.

Old Zeno files used a mixture of Latin1 and UTF-8 so there is still some "auto detection" code left in the zimlib, a workaround for this bug. This will be removed in future versions. Zeno files are not supported anymore.

Integer Encoding

For integer encoding the same algorithm as UTF-8 encoding is used. This encoding is also known as "integer compression". It safes some bytes by using variable lengths of integer fields, depending on the actual value of the number.

See also http://en.wikipedia.org/wiki/UTF-8#Design.

Old Zeno files used the QUnicode library instead. By switching to UTF-8 the new format is more standard-adherent and easier to understand.

Split ZIM files

ZIM files can be split in multiple chunks. This is necessary to be able to store big (over 4GB for example) ZIM files to limited file systems (like FAT32). That said, the chunks can be of any size, but the naming is really important. The ZIM file chunks should be named like following (the file name extensions matter): foobar.zimaa, foobar.zimab, foobar.zimac...

See also